Financing patterns of European SMEs – an empirical taxonomy

Abstract

This paper develops an empirical taxonomy of SME financing patterns in Europe by performing a cluster analysis including 12,726 SMEs in 28 European countries. The results reveal that SME financing in Europe is not homogenous but that different financing patterns exist. The cluster analysis identifies six distinct SME financing types: mixed-financed SMEs, state-subsidised SMEs, debt-financed SMEs, flexible-debt-financed SMEs, trade-financed SMEs and internally financed SMEs. These SME financing types differ according to the number of financing instruments used and the combinations thereof. Furthermore, the SME financing types can be profiled according to their firm-, product-, industry- and country-specific characteristics. Our findings support policy-makers in assessing the impact of policy changes on SME financing and in designing financing programmes tailored to the specific needs of SMEs.

Keywords:

• SME financing in Europe
• financing patterns
• empirical taxonomy
• cluster analysis

Acknowledgements

We would like to thank Dr Helmut Kraemer-Eis and Dr Frank Lang from the European Investment Fund (EIF) for their valuable suggestions and feedback. Furthermore, we would like to thank Annalisa Ferrando from the European Central Bank (ECB) for the fruitful discussions about our research project. An earlier version of this paper has been published as EIF working paper 2015/30.

Notes

1. To calculate the appropriate weights, the data on company size, economic activities and countries reported by Eurostat are used: http://appsso.eurostat.ec.europa.eu/nui/show.do?wai=true&data-set=sbs_sc_sca_r2 (accessed 15 December 2014).

2. The questionnaire is available at https://www.ecb.europa.eu/stats/money/surveys/sme/html/index.en.html (accessed 2 January 2016).

3. Basically, two statistical methods are appropriate to develop an empirical taxonomy: cluster analysis and correspondence analysis. As the result of a correspondence analysis would be rather complex and difficult to read as it would depict all 12,726 SMEs of our sub-sample in a two-dimensional space, we decided for the more appropriate cluster analysis (Backhaus et al. 2013).

4. However, we also applied other proximity measures to test for the stability of the clusters. The Rogers and Tanimoto as well as the Russel and Rao similarity measures produced a relatively high matching in the cluster solutions of 77.1 and 76.2%.

5. We validated the cluster results using the Test of Mojena and the Elbow Criterion (Mojena 1977; Backhaus et al. 2013). As both measures did not provide an unambiguous result, different cluster results were analysed and compared. This approach supported the six-cluster solution (Hair et al. 2010).

6. For more details about country distribution, please compare Table A1 (Appendix 1).

7. Due to the low relevance of debt securities issued and subordinated loans, participating loans, preferred stocks or similar financing instruments in the data-set, these groups were merged in the analysis into the category ‘Other’ (debt securities, subordinated/participating loans and preferred stock).

8. Table 4 should be read by comparing the share of SMEs per cluster and the share of SMEs in each category of passive cluster variables. For example, 32.5% of all SMEs with 1–9 employees are internally financed SMEs, even though only 31.4% of all SMEs belong to this cluster. This result suggests that smaller firms are more likely to be internally financed SMEs. Due to the large sample size, even small differences are noteworthy. The Pearson χ² test statistic of 120.8 for the variable ‘number of employees’ is statistically significant (p < 0.01). This result suggests that the differences in the number of employees are not due to chance.

9. Based on the SAFE survey, alternative instruments include trade credit, leasing, factoring and hire-purchase.

10. SMEs in the trade-financed SME cluster show more often than the average high to moderate past growth rates. Regarding innovation activity, they are in the average compared to the other clusters.

11. For a more detailed analysis, please compare Moritz (2015).

12. Due to a lack of data, Malta is not included in the analysis (see Section 3.1).

13. See https://www.ecb.europa.eu/ecb/orga/escb/html/mission_eurosys.en.html (accessed 2 January 2016).